RoboDojo-RealEval leaderboard

Real-robot track of RoboDojo with remote cloud access, standardized hardware and scene reset. Its tasks are not matched one-to-one with the simulation track.

Real robot · ARX X5, Piper, Piper X (bimanual)Report an error
33.9%Best · SLA
14Models ranked
OpenSaturation
Nov 18, 2026Numbers checked
#ModelTypeScoreDateSource
1SLAHKUST(GZ) & OLAVLA33.9%Nov 3, 2026
2InternW0-ΔShanghai AI LabWorld-action31.7%Oct 21, 2026
3MWA-PhoenixAnyverse & OpenPhoenixWorld-action30.0%Nov 18, 2026
4OpenWAM-αOpenWAM TeamWorld-action24.4%Sep 5, 2026
5π0.5Physical IntelligenceVLA12.8%, self-reported by the authors, not a leaderboard entryJul 2026
6InternVLA-A1Shanghai AI LabVLA7.2%, self-reported by the authors, not a leaderboard entryJul 2026
7GalaxeaVLA (G0)Galaxea AIVLA4.4%, self-reported by the authors, not a leaderboard entryJul 2026
8Xiaomi-Robotics-0Xiaomi RoboticsVLA3.9%, self-reported by the authors, not a leaderboard entryJul 2026
9X-VLATsinghua AIR et al.VLA3.3%, self-reported by the authors, not a leaderboard entryJul 2026
10GR00T N1.7NVIDIAVLA1.7%, self-reported by the authors, not a leaderboard entryJul 2026
10StarVLAStarVLA TeamVLA1.7%Jul 6, 2026
10π0Physical IntelligenceVLA1.7%, self-reported by the authors, not a leaderboard entryJul 2026
13Spirit v1.5Spirit AIVLA0.6%Jul 6, 2026
14DM0DexmalVLA0.0%Jul 6, 2026

Metric: Success rate over 18 real tasks (6 per platform), 10 trials each, scored by 3 blind evaluators. One result per model: RoboFYI re-run, then official leaderboard, then paper, then third party. Runs that deviate from the official protocol are listed on the model page but not ranked.

Wrong or missing number? Report an error